Exercises

Attempt one or two of the following exercises on your own:

1. Run the CrawlerStaticTarget example scene and compare its performance to the dynamic sample.

2. Double the time_horizon, batch_size, and buffer_size brain hyperparameters in one of the other control examples:

time_horizon: 2000
batch_size: 4048
buffer_size: 40480

3. Perform the same modification of time_horizon, batch_size, and buffer_size on another control sample and observe the combined effect.
4. Modify the num_layers and hidden_units brain hyperparameters to values we used in a control sample and apply them to a discrete action example, such as the Hallway example, as shown in the following code. How did it affect training?

num_layers: 3
hidden_units: 512

5. Alter the num_layers and hidden_units hyperparameters on another continuous or discrete action example and combine it with other parameter modifications.
6. Modify...